We Removed the Signup Wall and Let People Use the App First

The Setup

We had just finished a full teardown of a month of production data. The headline: our blog drew 11,001 sessions and converted at 0.19%, while our homepage converted at 13.6%. Activation for people who did sign up was a healthy 74%, and week-one retention was a grim 8%.

Two problems, then. Getting readers into the product, and getting users to come back. This post is about the first one.

Why "Add a Better CTA" Was the Wrong Answer

The obvious move when a page converts badly is to make the button better. Stronger verb, contrasting color, move it higher, add one above the fold and one below.

We had already done a round of that. It helped a little — CTR moved from 0.07% to 1.21% in the last week of July, which sounds enormous until you notice it is 1.21%.

The reason it stayed small is that the button was not the obstacle. What sat behind the button was the obstacle. Every marketing CTA deep-linked to the app with a signup parameter attached. Somebody reading an article about whiteboard apps, mildly curious, clicked through and got an account creation screen.

That is a terrible trade to offer a stranger. They have invested four minutes of reading. We are asking for an email address, a password, and a commitment, in exchange for a product they have not seen. The button was fine. The offer was bad.

The Change

We made the real app openable without an account.

Not a demo. Not a video. Not a stripped-down sandbox that hints at what the product might do. The actual application, with the actual canvas, opened directly from the article, with the toolbar live and the tools working.

Concretely:

• Marketing CTAs no longer force a signup mode parameter
• Article links can specify a template, so a whiteboarding article opens a whiteboard and a meeting-notes article opens the meeting-notes template
• Work is held locally until the user chooses to save it
• The account prompt appears after meaningful work exists, not before

The last point is where most of the design effort went.

Defining "Meaningful Work"

The prompt to create an account has to fire at the right moment. Too early and it is the signup wall again with extra steps. Too late and someone closes the tab with unsaved work, which is worse than never having engaged them.

Our first version counted objects and text length. Draw enough shapes or type enough characters and you are deemed to have done meaningful work.

That heuristic is weak, and we flagged it as the thing most likely to need revision. Object count rewards fidgeting. Someone who drags five rectangles around while deciding whether they like the tool registers as more invested than someone who wrote two sentences that actually matter to them. Character count has the mirror problem.

The direction we prefer is intent-shaped rather than volume-shaped: did this person do something that implies they want to keep it? Naming a note. Structuring content into a hierarchy. Recording audio. Returning to a canvas after navigating away. Those are cheap to detect and much harder to trigger by accident.

We shipped the volume heuristic because it was ready, and we wrote down why it is wrong so that the next version has somewhere to start.

The Second-Note Loop

The other half of the change targets retention rather than acquisition, and it comes from the same finding: the median new user creates exactly one note.

One note is a trial. Two notes is a beginning of a habit. So there is an intent-aware prompt that waits until someone has finished a first piece of real work, then suggests a specific next thing based on what they just did — not a generic "create another note" nudge, and not a tour.

The rules we gave it:

• Never fire before meaningful work exists
• Never fire more than once per session
• Suggest something concrete and related to what they just made
• Be dismissible permanently, in one click, with no follow-up

Anything that violates the last rule stops being onboarding and starts being nagging.

What Else Had to Change

Removing a signup wall touches more than the wall.

Positioning. The homepage led with features. It now leads with visual and spatial notes, because that is what the product is and it is what the highest-converting article was about.

A meeting-notes landing page. We ship a full AI meeting-notes suite and had no page for it. The AI-notes blog cluster had nowhere to send people.

Attribution that survives a subdomain hop. The marketing site and the app are different origins. Without a shared anonymous identity, every click from the blog into the app looked like direct traffic and every conversion was unattributable. We added a cross-subdomain anonymous identifier, first-touch attribution, signup joining, and a small set of privacy-minimal funnel events. Without this, we would ship the change and have no way to know whether it worked.

Lifecycle email deep links. The existing emails pointed at generic entry points. They now land where the message is about.

The Trap We Had to Handle

There is a specific failure mode in try-before-signup that quietly loses user data.

Somebody uses the product anonymously, makes something, then clicks a save or signup action. If the handoff into the authenticated account serializes only part of what they made, they will believe their work was saved when some of it was not.

In our case, canvas elements serialize cleanly, but things like recordings are sidecar data attached alongside the canvas rather than inside it. A naive handoff carries the drawing and silently drops the audio.

The rule we settled on: either the handoff carries everything, or the action is labeled precisely enough that nobody is surprised. "Save canvas" when only the canvas is going. Never "Save my work" when part of the work is going to disappear.

Losing a stranger's first ten minutes of work is the single most expensive bug in a try-before-signup flow, because that person never comes back and never tells you.

How We Will Know If It Worked

The before-state is measured and written down: 0.19% blog CTR, 74% activation, 17% day-one retention, 8% day-seven retention, 57 signups in the month.

The measurement plan is boring on purpose:

• Two clean weeks before drawing any conclusion
• Compare blog CTR against the homepage's 13.6%, which is the realistic ceiling
• Watch activation for anonymous users separately from account-first users
• Track the second-note rate as the leading retention indicator

We will publish the results either way. A post about a growth change that only appears when the change worked is not a data post, it is an advertisement.

Want to see what a blog reader now gets instead of a signup form? Open the canvas — no account.

30 days free. No credit card required.